Environment

• Python >= 3.7
• PyTorch >= 1.7
• NVIDIA GPU + CUDA

Installation

pip install -r requirements.txt
python setup.py develop

How To Inference or Test

• Refer to ./options/test for the configuration file of the model to be tested.
• Download the pretrained model from Google Drive and put it into ./experiments/UDCUNet_4gpu_pretrained/models .
• Prepare the testing data (validation data) into ./datasets/test/input (./datasets/validation/input and ./datasets/validation/GT)
• All datasets can be downloaded at the MIPI Challenage official website.
• Then run the follwing codes (Inference for example):

python basicsr/test.py -opt options/inference/UDCUNet_inference.yml

The testing results will be saved in the ./results folder.

How To Train

• Refer to ./options/train for the configuration file of the model to train.
• Prepare the training data in ./datasets/training/input and ./datasets/training/GT
• All datasets can be downloaded at the MIPI Challenage official website.
• The training command is like

Single GPU: CUDA_VISIBLE_DEVICES=0 python basicsr/train.py -opt options/train/UDCUNet_train.yml
Multi GPU: CUDA_VISIBLE_DEVICES=0,1,2,3 python -m torch.distributed.launch --nproc_per_node=4 --master_port=4321 basicsr/train.py -opt options/train/UDCUNet_train.yml --launcher pytorch

For more training commands and details, please check the docs in BasciSR

Contact

If you have any question, please email jf.hu1@siat.ac.cn.